5.1 Overview

Game Theory (GT) is well-known in economic studies as a theoretical approach to model the strategic interaction between multiple individuals or players in a specific situation. Game Theory approach can also be adopted in the science area to optimize decision-making processes in a strategic setting and has been successfully solved Multi-Agent RL (MARL) problems. If you would like to know more about the corporation between GT and MARL, you can have a look at these papers: (1) self-optimization using a State-based Potential Game approach and (2) potential game-based distributed optimization of a production unit.

MLPro-GT is developed as a sub-framework for the cooperative GT approach to solving MARL problems by inheriting a handful of main functionalities of MLPro-RL, such as the environment model, the agent model, and the environment-agent interaction model. MLPro-GT serves as a child package and MLPro-RL is its parent package, which is explained in the simplified diagram below:

You can easily access the GT module, as follows:

from mlpro.gt.models import *

You can check out some of the examples on our how to files or here.